Use 8192 bp final-position variant windows

README.md

@@ -80,7 +80,16 @@ Dataset-wide provenance summary:
     "single nucleotide variant": 6000
   },
   "window_bp": {
-    "1024": 6000
+    "8192": 6000
+  },
+  "variant_window_placement": {
+    "final": 6000
+  },
+  "variant_index_0based": {
+    "8191": 6000
+  },
+  "variant_position_1based": {
+    "8192": 6000
   },
   "benchmark_exclusion_status": {
     "checked": 6000
@@ -88,6 +97,8 @@ Dataset-wide provenance summary:
 }
 ```
 
+Every prompt sequence is 8,192 bp long. The variant is placed at the final base of the sequence context: `variant_index_0based = 8191`, equivalent to `variant_position_1based = 8192`.
+
 The `all_tasks` config has 6,000 task rows but 4,000 unique variant keys. The `binary` and `clinvar_levels` configs are two views of the same 2,000 clinical-classification variants; `pathogenic_pairwise` contributes 2,000 additional pathogenic or likely pathogenic variants.
 
 Unique variant chromosome distribution across `all_tasks`:
@@ -203,7 +214,7 @@ excluded_from_benchmark_eval: true
 
 ### `binary`
 
-The prompt shows the reference allele, alternate allele, and a 1,024 bp DNA sequence window containing the alternate allele. The model is asked to answer with exactly one option:
+The prompt shows the reference allele, alternate allele, and an 8,192 bp DNA sequence window containing the alternate allele at the final base of the context. The model is asked to answer with exactly one option:
 
 ```text
 benign or likely benign

all_tasks/train-00000-of-00001.jsonl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ad933e84128aae04f5ec0619689df861c1bb758aa7ceb7839f752063c4c792f8
-size 38219553
+oid sha256:e96535d15744930b81154e8d7c3ee96aff217f9c0a160070b7a9272739d3fe82
+size 210851553

binary/train-00000-of-00001.jsonl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:48707401850b0f51cb21555a63cb5ced83292cf34b915b57c9af71a7f30e4f7c
-size 10904557
+oid sha256:0192484cf7a48e60291f8a53d0160f61a67f393bf45d2f957e4fb72078f4ba71
+size 54112557

clinvar_levels/train-00000-of-00001.jsonl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dc14c14ed8b3a34485311f568ae3c225da95912fa947aec28e118b8d6b1838b7
-size 11240553
+oid sha256:a311857606a5e035bb4edb650307f4c3f9a62c39196be2973f350a396dd7fab1
+size 54448553

pathogenic_pairwise/train-00000-of-00001.jsonl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8a3ac9ee25ad5167fdeb49b3e9ab880144c245c1e1ecf2a87dd22a03c311abac
-size 16074443
+oid sha256:ad6a2ea8543d0a52c3798fe311da9914373150285f450ae515028bfd4f3b4233
+size 102290443